The present invention concerns the connection of a first electrical circuit to a second electrical circuit using a matching network so as to provide maximum power transfer between the first electrical circuit (the "source") and the second electrical circuit (the "load").
Maximum power is transferred from the source to the load when the output impedance of the source is the complex conjugate of the input impedance of the load. In most cases the output impedance of the source is not naturally equal to the complex conjugate of the input impedance of the load; therefore matching networks are placed between the source and load when power control and efficiency are critical. A matching network operates properly when the input impedance of the matching network is the complex conjugate of the output impedance of the source, and the output impedance of the matching network is the complex conjugate of the input impedance of the load. In this way power may be transferred from a source through a matching network to a load with minimal loss of power through power reflection, heat dissipation, etc.
In cases where the input impedance of the load varies during operation it is necessary to make adjustments to the matching network to maintain maximum power transfer from the source to the load. Typically, matching networks are designed such that variations in the input impedance of the load will result in a variation of the impedance of the matching network, the input impedance of the matching network being held constant. Further, in many applications the output impedance of a source is an output resistance with a negligible imaginary component. Therefore, in some prior art applications, the impedance magnitude and the impedance phase angle is measured at the input of the matching networks. Variable capacitors or inductors within the matching network are varied until the input impedance of the matching network matches the output impedance of the source network, that is until the impedance phase angle is zero and the impedance magnitude matches the magnitude of the output resistance of the source. The variable capacitors or inductors are placed in the matching network so that for every predicted variance in the input impedance of the load there is a solution in which the variable capacitors are set to values so that for the input of the matching network the impedance phase angle is zero and the impedance magnitude matches the magnitude of the output resistance of the source.
One problem with such prior art systems is that although there may be only one unique (or correct) solution in which the variable capacitors or inductors are set to values so that for the input of the matching network both the impedance phase angle is zero and the impedance magnitude matches the magnitude of the output resistance of the source; nevertheless, multiple solutions (i.e., false solutions) may exist where the variable capacitors are set to values so that for the input of the matching network either the impedance phase angle is zero or the impedance magnitude matches the magnitude of the output resistance of the source (but not both). This makes it difficult for a matching network to detect the unique or correct solution. The result may be that the matching network converges on a false solution, oscillates between multiple solutions or diverges completely.
In some applications, where the input of the impedance of the load does not vary significantly, it may be possible to initialize the variable capacitors or inductors to values which are close to the unique solution. This increases the likelihood that the matching network will converge on the correct solution.
Additionally, another method attempted in the prior art for adjusting the matching network to take into account variations in the input impedance of the load during operation is to measure reflected power at the input of the matching network. Under manual or computer control each of the variable capacitors or inductors, in turn, is varied while the other variable capacitors or inductors are held at constant values. By such iteration it is hoped to find a value for each variable capacitor so that the reflected power is as small as possible.
An additional problem with all the above-discussed prior art systems is that each solution relies on the use of variable capacitors or inductors in which capacitance is varied mechanically. In some applications, for example where the load impedance is generated by the application of current through a plasma used is semiconductor manufacturing processes, the use of mechanically variable capacitors may introduce significant delay into the process.